Spatial4D is hosted by MVP Lab for internal research use only. Access is granted case by case to lab members and approved collaborators. Every dataset in this repository keeps its original license and terms of use (see the dataset card); by requesting access you confirm that you have accepted the terms of each source dataset you will use, and that you will not redistribute the data.

Spatial4D

Spatial4D is MVP Lab's collection of 27 3D/4D datasets (multi-view images with depth, camera poses and intrinsics, and normals where available), converted to a common WebDataset layout for training feed-forward 3D reconstruction models. It totals 18.86 TB in 112,105 files.

Access and licensing

This repository is not a public release. MVP Lab hosts these copies for internal research use, behind manual access approval, and takes responsibility for complying with each source's terms. Nothing here grants any right beyond the original licenses: each dataset remains under its own license or terms of use, listed in Licenses and citations and in LICENSE.md. Several sources allow non-commercial use only, and some require every user to accept their terms directly with the provider. Before using a dataset, read and accept its terms. Do not redistribute any part of this repository.

Folder	Dataset	Size	Files	License / terms
`argoverse2_sensor`	Argoverse 2 Sensor Dataset	1.10 TB	852	CC BY-NC-SA 4.0 (data); MIT (code/APIs)
`arkitscenes`	ARKitScenes	0.58 TB	35,339	ARKitScenes LICENSE (Apple custom license: non-commercial grant plus conditional commercial grant)
`BlendedMVG`	BlendedMVS / BlendedMVG	0.23 TB	506	CC BY 4.0
`co3d`	Common Objects in 3D (CO3D v2)	1.82 TB	30,780	No dataset-specific license found; the official repo states 'The CO3D codebase is released under the CC BY-NC 4.0'
`dl3dv10k`	DL3DV-10K	1.88 TB	3,765	DL3DV-10K Terms of Use + CC BY-NC 4.0
`dynamic_replica`	Dynamic Replica (DynamicStereo)	1.91 TB	968	Dynamic Replica License Agreement (Meta click-through)
`hypersim`	Hypersim	89.0 GB	771	CC BY-SA 3.0 Unported
`mapfree`	Niantic Map-free Relocalization Dataset	0.15 TB	527	Niantic Map-free Relocalization Dataset License Agreement
`mapillary_depth`	Mapillary Planet-Scale Depth Dataset (MPSD)	0.16 TB	1,124	Not confirmed: no license text readable on the official pages
`matrix_city`	MatrixCity	0.20 TB	48	CC BY-NC 4.0
`megadepth`	MegaDepth	92.0 GB	272	CC BY 4.0 (depth maps, SfM models, etc.); original images under their own licenses
`megadepth_x`	MegaDepth-X	0.18 TB	1,879	CC BY 4.0 (depth maps, SfM models, reconstruction files); original images under their own licenses
`mvs_synth`	MVS-Synth	15.8 GB	122	No named license; stated restriction 'The data is for research and educational use only.'
`nuscenes`	nuScenes	74.1 GB	853	nuScenes Terms of Use - Non-Commercial (CC BY-NC-SA 4.0 with additional Dataset Terms)
`overmaps_1k_v2`	OverMaps-1K	1.25 TB	1,012	CC BY-NC 4.0 (gated HF repo)
`replica`	Replica Dataset	45.5 GB	25	Replica Dataset Research Terms
`scannetpp`	ScanNet++	1.87 TB	913	ScanNet++ Terms of Use
`scannetv2`	ScanNet v2	0.49 TB	1,517	ScanNet Terms of Use
`syn4d_rgbd`	Syn4D RGB-D (HF Syn4D/Syn4D_RGBD; part of Syn4D)	0.81 TB	4,686	No license on the downloaded repo (Syn4D/Syn4D_RGBD has no card/license tag); parent release Syn4D/Syn4D is CC BY 4.0 for dataset-authored components with third-party notices
`tartanair_v1`	TartanAir	0.32 TB	371	CC BY 4.0
`tartanair_v2`	TartanAir V2	1.00 TB	1,124	CC BY 4.0 (dataset); MIT (tartanairpy toolkit)
`taskonomy`	Taskonomy	2.04 TB	476	Taskonomy Dataset End User License Agreement (rev. Oct 21, 2021)
`texturepoorsfm`	TexturePoorSfM	0.6 GB	57	Dataset license not stated by the authors
`unrealstereo4k`	UnrealStereo4K	21.6 GB	11	MIT (tag on the author's HF dataset card; SMD-Nets repo LICENSE is MIT)
`vkitti2`	Virtual KITTI 2	19.1 GB	53	CC BY-NC-SA 3.0
`waymo_open_dataset`	Waymo Open Dataset (Perception)	0.74 TB	1,002	Waymo Dataset License Agreement for Non-Commercial Use (March 2025)
`wildrgbd`	WildRGB-D	1.76 TB	23,052	MIT (repo LICENSE, 'Copyright (c) 2024 rowdataset'); no dataset-specific license text

Sizes are decimal (1 TB = 10^12 bytes).

Layout

Each dataset keeps its folder name from our storage root (Data4D), so a download can be used directly as DATA4D_ROOT:

<dataset>/<webdataset folder>/
  <dataset>_<...>_<split>-NNNNNN.tar    # one or more scenes per shard
  shard_mapping.json                    # scene -> shard index
  dataset_info*.json                    # units, conventions, split source
<dataset>/splits/                       # scene lists per split, where the dataset has them

Inside a shard, each modality (images, depths, cameras, normals, masks, ...) is a nested tar whose k-th member is view k of the scene.

Folders with more than 9,000 shards are split into numbered subfolders of 5,000 shards to stay under the Hub's 10,000-entries-per-folder limit: co3d/webdataset_v2/train/000/, 001/, ..., wildrgbd/webdataset/000/, 001/, .... The FeedForward-Benchmark loaders find shards recursively and identify them by file name, so the split layout loads unchanged.

Three datasets are re-encoded versions of their originals, with the same content for training:

co3d/webdataset_v2: depth stored as float16 .npz, already cut by the loader's percentile threshold and zeroed outside image_mask & depth_mask.
overmaps_1k_v2/webdataset: unused groups (images_raw, masks_raw, gaussians, sparse) and frames without a COLMAP pose dropped; depth stored as thresholded float16 .npz.
taskonomy/webdataset_v2: each building's frames reordered along its view graph (reverse Cuthill-McKee); member bytes unchanged, frame_ids and the view graph permuted to match.

Download

hf download mvp-lab/Spatial4D --repo-type dataset --local-dir /path/to/Data4D \
  --include "hypersim/**"          # or omit --include for everything
export DATA4D_ROOT=/path/to/Data4D

Licenses and citations

Argoverse 2 Sensor Dataset (`argoverse2_sensor`)

Homepage: https://www.argoverse.org/av2.html
License / terms: CC BY-NC-SA 4.0 (data); MIT (code/APIs)
Conditions: Non-commercial only (examples given: no training models for a company's vehicle/product); attribution '(c) 2021 Argo AI, LLC'; ShareAlike.
Sharing: No separate redistribution clause; governed by CC BY-NC-SA 4.0 (sharing allowed non-commercially with attribution and same license).

Citation

@inproceedings{Argoverse2,
  author = {Benjamin Wilson and William Qi and Tanmay Agarwal and John Lambert and Jagjeet Singh and Siddhesh Khandelwal and Bowen Pan and Ratnesh Kumar and Andrew Hartnett and Jhony Kaesemodel Pontes and Deva Ramanan and Peter Carr and James Hays},
  title = {Argoverse 2: Next Generation Datasets for Self-driving Perception and Forecasting},
  booktitle = {Proceedings of the Neural Information Processing Systems Track on Datasets and Benchmarks (NeurIPS Datasets and Benchmarks 2021)},
  year = {2021}
}

ARKitScenes (`arkitscenes`)

Homepage: https://github.com/apple/ARKitScenes
License / terms: ARKitScenes LICENSE (Apple custom license: non-commercial grant plus conditional commercial grant)
Conditions: Personal, non-commercial, non-exclusive license to use/reproduce/modify/redistribute; a separate commercial grant applies only to licensees whose products had <700M monthly active users in every month before Aug 2024; no use of Apple marks; AS IS. README: 'Please refer to the LICENSE file for detailed information on using the dataset'; license questions to ARKitScenes-license@group.apple.com.
Sharing: Redistribution permitted ('use, reproduce, modify and redistribute ... for non-commercial purposes only'); unmodified full redistributions must retain the notice and disclaimers. Note the LICENSE text is written for 'Apple Software' but the README points to it for the dataset.

Citation

@inproceedings{dehghan2021arkitscenes,
  title = {{ARK}itScenes - A Diverse Real-World Dataset for 3D Indoor Scene Understanding Using Mobile {RGB}-D Data},
  author = {Gilad Baruch and Zhuoyuan Chen and Afshin Dehghan and Tal Dimry and Yuri Feigin and Peter Fu and Thomas Gebauer and Brandon Joffe and Daniel Kurz and Arik Schwartz and Elad Shulman},
  booktitle = {Thirty-fifth Conference on Neural Information Processing Systems Datasets and Benchmarks Track (Round 1)},
  year = {2021},
  url = {https://openreview.net/forum?id=tjZjv_qh_CE}
}

BlendedMVS / BlendedMVG (`BlendedMVG`)

Homepage: https://github.com/YoYo000/BlendedMVS
License / terms: CC BY 4.0
Conditions: Attribution required; commercial use and adaptation permitted. README: 'BlendedMVS and BlendedMVG are licensed under a Creative Commons Attribution 4.0 International License'.
Sharing: No explicit redistribution clause; CC BY 4.0 permits redistribution with attribution.

Citation

@article{yao2020blendedmvs,
  title = {BlendedMVS: A Large-scale Dataset for Generalized Multi-view Stereo Networks},
  author = {Yao, Yao and Luo, Zixin and Li, Shiwei and Zhang, Jingyang and Ren, Yufan and Zhou, Lei and Fang, Tian and Quan, Long},
  journal = {Computer Vision and Pattern Recognition (CVPR)},
  year = {2020}
}

Common Objects in 3D (CO3D v2) (`co3d`)

Homepage: https://ai.meta.com/datasets/co3d-dataset/
License / terms: No dataset-specific license found; the official repo states 'The CO3D codebase is released under the CC BY-NC 4.0'
Conditions: Repo LICENSE is the full CC BY-NC 4.0 text (non-commercial, attribution), but the README scopes it to the codebase. Meta's dataset and download pages (ai.meta.com/datasets/co3d-dataset/, /co3d-downloads/) show no readable license text (download links are behind a Meta login).
Sharing: No explicit statement about redistributing the data found on official pages.

Citation

@inproceedings{reizenstein21co3d,
  author = {Reizenstein, Jeremy and Shapovalov, Roman and Henzler, Philipp and Sbordone, Luca and Labatut, Patrick and Novotny, David},
  booktitle = {International Conference on Computer Vision},
  title = {Common Objects in 3D: Large-Scale Learning and Evaluation of Real-life 3D Category Reconstruction},
  year = {2021}
}

DL3DV-10K (`dl3dv10k`)

Homepage: https://dl3dv-10k.github.io/DL3DV-10K/
License / terms: DL3DV-10K Terms of Use + CC BY-NC 4.0
Conditions: Non-commercial research and educational use only; access via gated Hugging Face request (requesting access = agreeing to terms); CC BY-NC 4.0 attribution. ToU PDF: https://github.com/DL3DV-10K/Dataset/blob/main/DL3DV-10K_term_of_use.pdf
Sharing: Researcher may give access to research associates/colleagues only if they first agree to the terms; 'Sharing the data otherwise is strictly prohibited.'

Citation

@inproceedings{ling2024dl3dv,
  title = {Dl3dv-10k: A large-scale scene dataset for deep learning-based 3d vision},
  author = {Ling, Lu and Sheng, Yichen and Tu, Zhi and Zhao, Wentian and Xin, Cheng and Wan, Kun and Yu, Lantao and Guo, Qianyu and Yu, Zixun and Lu, Yawen and others},
  booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  pages = {22160--22169},
  year = {2024}
}

Dynamic Replica (DynamicStereo) (`dynamic_replica`)

Homepage: https://dynamic-stereo.github.io/
License / terms: Dynamic Replica License Agreement (Meta click-through)
Conditions: Use only for research, educational, or artistic purposes that are non-commercial/not-for-profit; employees of for-profit entities may use it for research (employer also bound); AS IS; California law. Code: mostly CC BY-NC (github.com/facebookresearch/dynamic_stereo/blob/main/LICENSE).
Sharing: 'Researcher may provide research associates and colleagues with access to the Dataset provided that they first agree to be bound by these terms and conditions.' No other redistribution right granted.

Citation

@article{karaev2023dynamicstereo,
  title = {DynamicStereo: Consistent Dynamic Depth from Stereo Videos},
  author = {Nikita Karaev and Ignacio Rocco and Benjamin Graham and Natalia Neverova and Andrea Vedaldi and Christian Rupprecht},
  journal = {CVPR},
  year = {2023}
}

Hypersim (`hypersim`)

Homepage: https://github.com/apple/ml-hypersim
License / terms: CC BY-SA 3.0 Unported
Conditions: Attribution + ShareAlike; commercial use permitted by the license. README: 'The Hypersim Dataset is licensed under the Creative Commons Attribution-ShareAlike 3.0 Unported License.' Ground-truth meshes are NOT included (must purchase Evermotion assets via TurboSquid). Toolkit code under repo LICENSE.txt.
Sharing: No explicit redistribution clause; CC BY-SA 3.0 permits redistribution with attribution under the same license.

Citation

@inproceedings{roberts:2021,
  author = {Mike Roberts AND Jason Ramapuram AND Anurag Ranjan AND Atulit Kumar AND Miguel Angel Bautista AND Nathan Paczan AND Russ Webb AND Joshua M. Susskind},
  title = {{Hypersim}: {A} Photorealistic Synthetic Dataset for Holistic Indoor Scene Understanding},
  booktitle = {International Conference on Computer Vision (ICCV) 2021},
  year = {2021}
}

Niantic Map-free Relocalization Dataset (`mapfree`)

Homepage: https://www.nianticspatial.com/research/map-free
License / terms: Niantic Map-free Relocalization Dataset License Agreement
Conditions: Non-commercial only (academic research, teaching, demos, personal experimentation; publishing comparison results incl. retraining allowed); agreement terminates immediately upon commercial use; for-profit employer also bound; Niantic may revoke access; indemnity.
Sharing: No explicit redistribution clause; grant is a 'non-exclusive, non-transferable, non-sublicensable right to download and use' the dataset ('licensed, not sold').

Citation

@inproceedings{arnold2022mapfree,
  title = {Map-free Visual Relocalization: Metric Pose Relative to a Single Image},
  author = {Arnold, Eduardo and Wynn, Jamie and Vicente, Sara and Garcia-Hernando, Guillermo and Monszpart, {\'{A}}ron and Prisacariu, Victor Adrian and Turmukhambetov, Daniyar and Brachmann, Eric},
  booktitle = {ECCV},
  year = {2022}
}

Mapillary Planet-Scale Depth Dataset (MPSD) (`mapillary_depth`)

Homepage: https://www.mapillary.com/dataset/depth
License / terms: Not confirmed: no license text readable on the official pages
Conditions: The official dataset page is a login-gated JS app; its terms could not be retrieved. The ECCV 2020 paper and the Mapillary blog announcement state no license. A third-party paper (MapAnything, arXiv 2509.13414) lists MPSD as CC BY-NC-SA (unverified against Mapillary). Sibling Mapillary Vistas is CC BY-NC-SA plus a research-use agreement, which does not confirm MPSD's terms. Imagery is derived from Mapillary's public image database.
Sharing: None found on official pages.

Citation

@inproceedings{lopez2020mapillary,
  title = {Mapillary Planet-Scale Depth Dataset},
  author = {L{\'o}pez-Antequera, Manuel and Gargallo, Pau and Hofinger, Markus and Rota Bul{\`o}, Samuel and Kuang, Yubin and Kontschieder, Peter},
  booktitle = {European Conference on Computer Vision (ECCV)},
  year = {2020}
}

MatrixCity (`matrix_city`)

Homepage: https://city-super.github.io/matrixcity/
License / terms: CC BY-NC 4.0
Conditions: Non-commercial; attribution required; project page states the dataset is 'copyright by us'.
Sharing: No explicit redistribution clause; CC BY-NC 4.0 permits non-commercial sharing with attribution.

Citation

@inproceedings{li2023matrixcity,
  title = {Matrixcity: A large-scale city dataset for city-scale neural rendering and beyond},
  author = {Li, Yixuan and Jiang, Lihan and Xu, Linning and Xiangli, Yuanbo and Wang, Zhenzhi and Lin, Dahua and Dai, Bo},
  booktitle = {Proceedings of the IEEE/CVF International Conference on Computer Vision},
  pages = {3205--3215},
  year = {2023}
}

MegaDepth (`megadepth`)

Homepage: https://research.cs.cornell.edu/megadepth/
License / terms: CC BY 4.0 (depth maps, SfM models, etc.); original images under their own licenses
Conditions: Attribution; commercial use allowed for MegaDepth-produced assets. 'The original images come with their own licenses.' (Flickr/Internet photos).
Sharing: No explicit redistribution clause; CC BY 4.0 permits redistribution of the derived assets; images carry their own (per-image) licenses.

Citation

@inproceedings{MegaDepthLi18,
  title = {MegaDepth: Learning Single-View Depth Prediction from Internet Photos},
  author = {Zhengqi Li and Noah Snavely},
  booktitle = {Computer Vision and Pattern Recognition (CVPR)},
  year = {2018}
}

MegaDepth-X (`megadepth_x`)

Homepage: https://megadepth-x.github.io/
License / terms: CC BY 4.0 (depth maps, SfM models, reconstruction files); original images under their own licenses
Conditions: Attribution. Card: assets 'are licensed under the Creative Commons Attribution 4.0 International License. The original images come with their own licenses.' Some images were removed to comply with original image licenses. HF repo not gated; no YAML license tag. Built from MegaScenes (cite both).
Sharing: 'We encourage users to verify image licenses themselves before further redistribution.'

Citation

@inproceedings{li2026longtail,
  title = {Long-Tail Internet Photo Reconstruction},
  author = {Li, Yuan and Xiangli, Yuanbo and Averbuch-Elor, Hadar and Snavely, Noah and Cai, Ruojin},
  booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  year = {2026}
}
@inproceedings{tung2024megascenes,
  title = {MegaScenes: Scene-Level View Synthesis at Scale},
  author = {Tung, Joseph and Chou, Gene and Cai, Ruojin and Yang, Guandao and Zhang, Kai and Wetzstein, Gordon and Hariharan, Bharath and Snavely, Noah},
  booktitle = {Proceedings of the European Conference on Computer Vision},
  year = {2024}
}

MVS-Synth (`mvs_synth`)

Homepage: https://phuang17.github.io/DeepMVS/mvs-synth.html
License / terms: No named license; stated restriction 'The data is for research and educational use only.'
Conditions: Research and educational use only. Frames captured in Grand Theft Auto V (Rockstar Games content). Official HF repo phuang17/MVS-Synth tags license 'other' with the same sentence.
Sharing: No statement on redistribution.

Citation

@inproceedings{DeepMVS,
  author = {Huang, Po-Han and Matzen, Kevin and Kopf, Johannes and Ahuja, Narendra and Huang, Jia-Bin},
  title = {DeepMVS: Learning Multi-View Stereopsis},
  booktitle = {IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
  year = {2018}
}

nuScenes (`nuscenes`)

Homepage: https://www.nuscenes.org/nuscenes
License / terms: nuScenes Terms of Use - Non-Commercial (CC BY-NC-SA 4.0 with additional Dataset Terms)
Conditions: Non-commercial; datasets 'provided ... under a CC BY-NC-SA 4.0 ... with the additional terms included herein' (Dataset Terms prevail on conflict; last updated Nov 16, 2021); no implying Motional endorsement; Motional may terminate access; indemnity. Commercial use needs separate license (https://www.nuscenes.org/terms-of-use-commercial).
Sharing: No general redistribution ban (CC BY-NC-SA 4.0 sharing terms apply), but: 'Some datasets and indicators are provided by third parties and may not be redistributed or reused without the consent of the original data provider'.

Citation

@inproceedings{caesar2020nuscenes,
  title = {nuScenes: A multimodal dataset for autonomous driving},
  author = {Caesar, Holger and Bankiti, Varun and Lang, Alex H. and Vora, Sourabh and Liong, Venice Erin and Xu, Qiang and Krishnan, Anush and Pan, Yu and Baldan, Giancarlo and Beijbom, Oscar},
  booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
  year = {2020}
}

OverMaps-1K (`overmaps_1k_v2`)

Homepage: https://huggingface.co/datasets/OverTheReality/OverMaps_1k
License / terms: CC BY-NC 4.0 (gated HF repo)
Conditions: Research use allowed, commercial use prohibited; gated (must accept conditions to access files); free research-only 1,000-sample subset of a 155k commercial collection (data@ovr.ai); users responsible for privacy compliance.
Sharing: 'Please attribute OverTheReality and link back to this repository when sharing derivatives.'

Citation

@misc{OverMaps1k,
  author = {OverTheReality},
  title = {{OverMaps-1K Dataset}},
  howpublished = {Hugging Face Datasets},
  url = {https://huggingface.co/datasets/OverTheReality/OverMaps_1k},
  year = {2025}
}

Replica Dataset (`replica`)

Homepage: https://github.com/facebookresearch/Replica-Dataset
License / terms: Replica Dataset Research Terms
Conditions: Use only for research or educational purposes that are non-commercial/not-for-profit; employees of for-profit entities may use it under the same limits (employer bound); AS IS; California law.
Sharing: 'Researcher may provide research associates and colleagues with access to the Dataset provided that they first agree to be bound by these terms and conditions.' No other redistribution right granted.

Citation

@article{replica19arxiv,
  title = {The {R}eplica Dataset: A Digital Replica of Indoor Spaces},
  author = {Julian Straub and Thomas Whelan and Lingni Ma and Yufan Chen and Erik Wijmans and Simon Green and Jakob J. Engel and Raul Mur-Artal and Carl Ren and Shobhit Verma and Anton Clarkson and Mingfei Yan and Brian Budge and Yajie Yan and Xiaqing Pan and June Yon and Yuyang Zou and Kimberly Leon and Nigel Carter and Jesus Briales and Tyler Gillingham and Elias Mueggler and Luis Pesqueira and Manolis Savva and Dhruv Batra and Hauke M. Strasdat and Renzo De Nardi and Michael Goesele and Steven Lovegrove and Richard Newcombe},
  journal = {arXiv preprint arXiv:1906.05797},
  year = {2019}
}

ScanNet++ (`scannetpp`)

Homepage: https://scannetpp.mlsg.cit.tum.de/scannetpp/
License / terms: ScanNet++ Terms of Use
Conditions: Signed data-use agreement with TUM; non-commercial research/educational use only.
Sharing: No redistribution without permission from TUM (per existing FeedForward-Benchmark registry entry).

Citation

@inproceedings{yeshwanth2023scannetpp,
  title = {ScanNet++: A High-Fidelity Dataset of 3D Indoor Scenes},
  author = {Yeshwanth, Chandan and Liu, Yueh-Cheng and Nie{\ss}ner, Matthias and Dai, Angela},
  booktitle = {Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV)},
  year = {2023}
}

ScanNet v2 (`scannetv2`)

Homepage: http://www.scan-net.org/
License / terms: ScanNet Terms of Use
Conditions: Signed terms of use; non-commercial research and educational use only.
Sharing: Terms do not permit redistribution of the data (per existing FeedForward-Benchmark registry entry, status no_redistribution).

Citation

@inproceedings{dai2017scannet,
  title = {ScanNet: Richly-annotated 3D Reconstructions of Indoor Scenes},
  author = {Dai, Angela and Chang, Angel X. and Savva, Manolis and Halber, Maciej and Funkhouser, Thomas and Nie{\ss}ner, Matthias},
  booktitle = {Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
  year = {2017}
}

Syn4D RGB-D (HF Syn4D/Syn4D_RGBD; part of Syn4D) (`syn4d_rgbd`)

Homepage: https://jzr99.github.io/Syn4D
License / terms: No license on the downloaded repo (Syn4D/Syn4D_RGBD has no card/license tag); parent release Syn4D/Syn4D is CC BY 4.0 for dataset-authored components with third-party notices
Conditions: Parent Syn4D/Syn4D: CC BY 4.0 for dataset-authored components; third-party assets keep their licenses (Fab Standard License for Unreal scenes, Objaverse per-object licenses, Google Scanned Objects CC BY 4.0, BEDLAM2 not redistributed - users must obtain it and accept its license). Source of our copy: Syn4D/Syn4D_RGBD (download_manifest.json), whose HF discussion #2 'what is the dataset license?' (Jun 2026) is unanswered.
Sharing: Parent card: 'Processed BEDLAM2 metadata is not redistributed due to BEDLAM2 license restrictions'; Fab source assets not redistributed. Nothing stated on Syn4D_RGBD itself.

Citation

@misc{jiang2026syn4d,
  title = {Syn4D: A Multiview Synthetic 4D Dataset},
  author = {Jiang, Zeren and Lan, Yushi and Luo, Yihang and Deng, Yufan and Lai, Zihang and Sucar, Edgar and Rupprecht, Christian and Laina, Iro and Larlus, Diane and Zheng, Chuanxia and Vedaldi, Andrea},
  year = {2026},
  eprint = {2605.05207},
  archivePrefix = {arXiv},
  primaryClass = {cs.CV},
  url = {https://arxiv.org/abs/2605.05207}
}

TartanAir (`tartanair_v1`)

Homepage: https://theairlab.org/tartanair-dataset/
License / terms: CC BY 4.0
Conditions: Attribution; commercial use permitted. Stated under the page's 'Term of use' heading: 'This work is licensed under a Creative Commons Attribution 4.0 International License.'
Sharing: No explicit redistribution clause; CC BY 4.0 permits redistribution with attribution.

Citation

@inproceedings{tartanair2020iros,
  title = {TartanAir: A Dataset to Push the Limits of Visual SLAM},
  author = {Wang, Wenshan and Zhu, Delong and Wang, Xiangwei and Hu, Yaoyu and Qiu, Yuheng and Wang, Chen and Hu, Yafei and Kapoor, Ashish and Scherer, Sebastian},
  booktitle = {2020 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)},
  year = {2020}
}

TartanAir V2 (`tartanair_v2`)

Homepage: https://tartanair.org/
License / terms: CC BY 4.0 (dataset); MIT (tartanairpy toolkit)
Conditions: Attribution; commercial use permitted. Copyright Carnegie Mellon University 2023 (Wenshan Wang, Yaoyu Hu, Yuheng Qiu, Shihao Shen, Yorai Shaoul).
Sharing: No explicit redistribution clause; CC BY 4.0 permits redistribution with attribution.

Citation

@inproceedings{tartanair2020iros,
  title = {TartanAir: A Dataset to Push the Limits of Visual SLAM},
  author = {Wang, Wenshan and Zhu, Delong and Wang, Xiangwei and Hu, Yaoyu and Qiu, Yuheng and Wang, Chen and Hu, Yafei and Kapoor, Ashish and Scherer, Sebastian},
  booktitle = {2020 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)},
  year = {2020}
}
@inproceedings{patel2025tartanground,
  title = {TartanGround: A Large-Scale Dataset for Ground Robot Perception and Navigation},
  author = {Patel, Manthan and Yang, Fan and Qiu, Yuheng and Cadena, Cesar and Scherer, Sebastian and Hutter, Marco and Wang, Wenshan},
  booktitle = {IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)},
  year = {2025}
}

Taskonomy (`taskonomy`)

Homepage: http://taskonomy.stanford.edu/
License / terms: Taskonomy Dataset End User License Agreement (rev. Oct 21, 2021)
Conditions: Research/educational use that is non-commercial or not-for-profit only (commercial research orgs may use for research); must reference Taskonomy Dataset or the agreement when publishing derived information.
Sharing: 'The data may not be redistributed, in whole or part, or in any format without written prior permission.' New access must go through the official form/distribution method.

Citation

@inproceedings{zamir2018taskonomy,
  title = {Taskonomy: Disentangling Task Transfer Learning},
  author = {Zamir, Amir R and Sax, Alexander and Shen, William B and Guibas, Leonidas and Malik, Jitendra and Savarese, Silvio},
  booktitle = {2018 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
  year = {2018},
  organization = {IEEE}
}

TexturePoorSfM (`texturepoorsfm`)

Homepage: https://github.com/zju3dv/DetectorFreeSfM
License / terms: Dataset license not stated by the authors
Conditions: No data license published; DetectorFreeSfM code license (Apache-2.0, later PRL v1.0) does not cover the data.
Sharing: None stated.

Citation

@inproceedings{he2024detectorfree,
  title = {Detector-Free Structure from Motion},
  author = {He, Xingyi and Sun, Jiaming and Wang, Yifan and Peng, Sida and Huang, Qixing and Bao, Hujun and Zhou, Xiaowei},
  booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
  year = {2024}
}

UnrealStereo4K (`unrealstereo4k`)

Homepage: https://github.com/fabiotosi92/SMD-Nets
License / terms: MIT (tag on the author's HF dataset card; SMD-Nets repo LICENSE is MIT)
Conditions: Author's HF card (quarter-res release) tags 'License: mit' and says 'This dataset maintains the same license as the original UnrealStereo4K dataset', pointing to the SMD-Nets repo, whose MIT LICENSE does not explicitly say whether it covers the data. No other dataset terms stated.
Sharing: No explicit statement; MIT would permit redistribution with the notice.

Citation

@inproceedings{Tosi2021CVPR,
  author = {Fabio Tosi and Yiyi Liao and Carolin Schmitt and Andreas Geiger},
  title = {SMD-Nets: Stereo Mixture Density Networks},
  booktitle = {Conference on Computer Vision and Pattern Recognition (CVPR)},
  year = {2021}
}

Virtual KITTI 2 (`vkitti2`)

Homepage: https://europe.naverlabs.com/research/computer-vision/proxy-virtual-worlds-vkitti-2/
License / terms: CC BY-NC-SA 3.0
Conditions: Non-commercial only; attribution (cite papers and Naver as originator); ShareAlike; copyright Naver Corporation.
Sharing: No explicit redistribution clause; CC BY-NC-SA 3.0 permits non-commercial sharing with attribution under the same license.

Citation

@misc{cabon2020vkitti2,
  title = {Virtual KITTI 2},
  author = {Cabon, Yohann and Murray, Naila and Humenberger, Martin},
  year = {2020},
  eprint = {2001.10773},
  archivePrefix = {arXiv},
  primaryClass = {cs.CV}
}
@inproceedings{gaidon2016virtual,
  title = {Virtual worlds as proxy for multi-object tracking analysis},
  author = {Gaidon, Adrien and Wang, Qiao and Cabon, Yohann and Vig, Eleonora},
  booktitle = {Proceedings of the IEEE conference on Computer Vision and Pattern Recognition},
  pages = {4340--4349},
  year = {2016}
}

Waymo Open Dataset (Perception) (`waymo_open_dataset`)

Homepage: https://waymo.com/open/
License / terms: Waymo Dataset License Agreement for Non-Commercial Use (March 2025)
Conditions: Non-commercial only; must not use/deploy the dataset or models trained on it in vehicles, production systems or for primarily commercial purposes; personal, non-exclusive license; commercial license separately (open-dataset@waymo.com); on non-compliance must delete the dataset.
Sharing: Section 2: 'You may distribute the Dataset and copies and modifications thereof, but only if' distributed to individuals who have registered and agreed to the Terms; small extracts allowed for illustration in publications; trained models may be published with a copy of the Agreement.

Citation

@inproceedings{Sun_2020_CVPR,
  author = {Sun, Pei and Kretzschmar, Henrik and Dotiwalla, Xerxes and Chouard, Aurelien and Patnaik, Vijaysai and Tsui, Paul and Guo, James and Zhou, Yin and Chai, Yuning and Caine, Benjamin and Vasudevan, Vijay and Han, Wei and Ngiam, Jiquan and Zhao, Hang and Timofeev, Aleksei and Ettinger, Scott and Krivokon, Maxim and Gao, Amy and Joshi, Aditya and Zhang, Yu and Shlens, Jonathon and Chen, Zhifeng and Anguelov, Dragomir},
  title = {Scalability in Perception for Autonomous Driving: Waymo Open Dataset},
  booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
  month = {June},
  year = {2020}
}

WildRGB-D (`wildrgbd`)

Homepage: https://wildrgbd.github.io/
License / terms: MIT (repo LICENSE, 'Copyright (c) 2024 rowdataset'); no dataset-specific license text
Conditions: The official dataset repo carries an MIT LICENSE, but its text covers 'Software' and does not mention data; the project page states no data license.
Sharing: No explicit statement about the data.

Citation

@misc{xia2024rgbd,
  title = {RGBD Objects in the Wild: Scaling Real-World 3D Object Learning from RGB-D Videos},
  author = {Hongchi Xia and Yang Fu and Sifei Liu and Xiaolong Wang},
  year = {2024},
  eprint = {2401.12592},
  archivePrefix = {arXiv},
  primaryClass = {cs.CV}
}

Downloads last month: 207

Total file size:

13.4 TB

Papers for mvp-lab/Spatial4D

Syn4D: A Multiview Synthetic 4D Dataset

Paper • 2605.05207 • Published May 6 • 1

RGBD Objects in the Wild: Scaling Real-World 3D Object Learning from RGB-D Videos

Paper • 2401.12592 • Published Jul 28, 2024

Virtual KITTI 2

Paper • 2001.10773 • Published Jan 29, 2020

The Replica Dataset: A Digital Replica of Indoor Spaces

Paper • 1906.05797 • Published Jun 13, 2019

You need to agree to share your Hugging Face information to access this dataset

Spatial4D

Access and licensing

Contents

Layout

Download

Licenses and citations

Argoverse 2 Sensor Dataset (argoverse2_sensor)

ARKitScenes (arkitscenes)

BlendedMVS / BlendedMVG (BlendedMVG)

Common Objects in 3D (CO3D v2) (co3d)

DL3DV-10K (dl3dv10k)

Dynamic Replica (DynamicStereo) (dynamic_replica)

Hypersim (hypersim)

Niantic Map-free Relocalization Dataset (mapfree)

Mapillary Planet-Scale Depth Dataset (MPSD) (mapillary_depth)

MatrixCity (matrix_city)

MegaDepth (megadepth)

MegaDepth-X (megadepth_x)

MVS-Synth (mvs_synth)

nuScenes (nuscenes)

OverMaps-1K (overmaps_1k_v2)

Replica Dataset (replica)

ScanNet++ (scannetpp)

ScanNet v2 (scannetv2)

Syn4D RGB-D (HF Syn4D/Syn4D_RGBD; part of Syn4D) (syn4d_rgbd)

TartanAir (tartanair_v1)

TartanAir V2 (tartanair_v2)

Taskonomy (taskonomy)

TexturePoorSfM (texturepoorsfm)

UnrealStereo4K (unrealstereo4k)

Virtual KITTI 2 (vkitti2)

Waymo Open Dataset (Perception) (waymo_open_dataset)

WildRGB-D (wildrgbd)

Papers for mvp-lab/Spatial4D

Argoverse 2 Sensor Dataset (`argoverse2_sensor`)

ARKitScenes (`arkitscenes`)

BlendedMVS / BlendedMVG (`BlendedMVG`)

Common Objects in 3D (CO3D v2) (`co3d`)

DL3DV-10K (`dl3dv10k`)

Dynamic Replica (DynamicStereo) (`dynamic_replica`)

Hypersim (`hypersim`)

Niantic Map-free Relocalization Dataset (`mapfree`)

Mapillary Planet-Scale Depth Dataset (MPSD) (`mapillary_depth`)

MatrixCity (`matrix_city`)

MegaDepth (`megadepth`)

MegaDepth-X (`megadepth_x`)

MVS-Synth (`mvs_synth`)

nuScenes (`nuscenes`)

OverMaps-1K (`overmaps_1k_v2`)

Replica Dataset (`replica`)

ScanNet++ (`scannetpp`)

ScanNet v2 (`scannetv2`)

Syn4D RGB-D (HF Syn4D/Syn4D_RGBD; part of Syn4D) (`syn4d_rgbd`)

TartanAir (`tartanair_v1`)

TartanAir V2 (`tartanair_v2`)

Taskonomy (`taskonomy`)

TexturePoorSfM (`texturepoorsfm`)

UnrealStereo4K (`unrealstereo4k`)

Virtual KITTI 2 (`vkitti2`)

Waymo Open Dataset (Perception) (`waymo_open_dataset`)

WildRGB-D (`wildrgbd`)